Research On The Influencing Factors Of The Hits On Product Link In The Microblogging Of Web Celebrity

From 2016 to the present,the web celebrity has become an increasingly important part of the Internet economy.The revenue of Chinese web celebrities reached 52.8 billion yuan in 2016 and will be more than 100 billion yuan in 2018.The compound annual growth rate is 59.4%from 2015 to 2018.In the context of consumption upgrade and social network e-commerce becomes the mainstream,e-commerce will be the most popular and matural way of cashing.In this paper,we made corresponding theoretical analyses according to the development process of web celebrity and the fact that they take the advantage of microblog to attract consumers to buy cloths in their online taobao shop.In addition,we studied the effect of the microblog marketing.This paper provides scientific method and basis for the cashing of the fans traffic of web celebrity and theoretical basis for the future research.The main contents of this paper include the following two aspects:firstly,to discuss the importance ranking of the factors influencing the hits of product links in the micro-blog of web celebrities.Secondly,to discuss the qualitative influence of important features.First of all,we summarized the development process and the current situation of web celebrity.We use the open source crawler framework Scrapy combining with the open-source browser engine Webkit technology to collect 221615 pieces of microblog data of 130 web celebrities.Then,we filter the data according to taobao domain name getting a total of 67306 pieces of microblog data including data that can measure personal influence(number of concerns,number of fans,number of microblogs released,whether certified),the data of microblogs released(release time,the length of content,the number of pictures contained,tag,emotion,originality)and the data related to fans' behavior(number of comments,number of highly-like comments,number of bloggers' reply,number of like,number of forward).Then we use the random forest algorithm to get the order of feature importance that affects the hits on product links.Then,regression analysis of important characteristics is made to qualitatively describes how these features affect hits.By analyzing the data,we find that the data of microblogs released and fans'behavior have a significant effect on the hits of product link in microblogging.The top five features are number of comments,the length of content,the number of pictures contained,number of forward,number of like.This shows that the content of microblogging and fans interaction are fundamental.According to the results of the regression analysis of random forest,the click volume is in proportional to the number of comments,the number of forwards,the number of likes;is inversely proportional to the length of the microblogging content;is not related to the number of pictures;hits of microblogging released in the evening and early hours are slightly higher than the microblogging released during the day;hits of original microblogging and tagged microblogging are slightly higher than the copy content and microblogging without tagging.The conclusion drawn in this paper provides recommendations to the promotion of web celebrity e-commerce and help web celebrities make better use of the microblog to improve sales and promote the development of the web celebrity industry.At the same time this information can also help future researchers.